Download Isotropic diffusion weighting in radial fast spin

Magnetic Resonance in Medicine 53:1347–1354 (2005)
Isotropic Diffusion Weighting in Radial Fast Spin-Echo
Magnetic Resonance Imaging
Joelle E. Sarlls,1 Rexford D. Newbould,1,2 Maria I. Altbach,3 Arthur F. Gmitro,1,3,4
Joachim Seeger,3 and Theodore P. Trouard1,3*
Radial fast spin-echo (radial-FSE) methods enable multishot
diffusion-weighted MRI (DWMRI) to be carried out without significant artifacts due to motion and/or susceptibility and can be
used to generate DWMRI images with high spatial resolution. In
this work, a novel method that allows isotropic diffusion
weighting to be obtained in a single radial k-space data set is
presented. This is accomplished by altering the direction of
diffusion weighting gradients between groups of TR periods,
which yield sets of radial lines that possess diffusion weighting
sensitive to motion in different directions. By altering the diffusion weighting directions and controlling the view ordering appropriately within the sequence, an effectively isotropic diffusion-weighted image can be obtained within one radial-FSE
scan. The order in which radial lines are acquired can also be
controlled to yield data sets without significant artifacts due to
motion, T2 decay, and/or diffusion anisotropy. Magn Reson
Med 53:1347–1354, 2005. © 2005 Wiley-Liss, Inc.
Key words: diffusion; ADC; radial; projection reconstruction;
MRI; stroke
Diffusion-weighted MRI (DWMRI) has become a powerful
tool for investigating the translational motion of water
within the human body. Because this motion is sensitive
to the cellular architecture and integrity of the tissue,
DWMRI has been used to investigate a number of diseases
including stroke, cancer, and many other neurologic disorders (1–5). Many tissues, particularly in the brain, exhibit long-range structural order, which imparts anisotropy to the translational motion of water. This feature has
been exploited in diffusion tensor imaging to investigate
the integrity of white matter structures in the brain and to
map out fiber orientations (6). In some applications, however, it is desirable to measure only the mean or average
diffusivity, without the effects of anisotropy. This is the
case in DWMRI of acute stroke where change in the apparent diffusion coefficient (ADC) of water is indicative of
ischemic tissue. In such situations, anisotropy from organized structures can cause errors in calculations of ADC if
diffusion is only measured in a single direction. To accu-
1
Biomedical Engineering Program, University of Arizona, Tuscon, Arizona,
USA.
2
Department of Electrical and Computer Engineering, University of Arizona,
Tucson, Arizona, USA.
3
Department of Radiology, University of Arizona, Tucson, Arizona, USA.
4
Optical Sciences Center, University of Arizona, Tucson, Arizona, USA.
Grant sponsor: National Institutes of Health; Grant number: R21 AG021624;
Grant sponsor: Flinn Foundation.
*Correspondence to: Theodore P. Trouard, Biomedical Engineering Program,
P.O. Box 245084, University of Arizona, Tucson, AZ 85724-5084, USA.
E-mail: [email protected]
Received 13 July 2004; revised 20 December 2004; accepted 29 December
2004.
DOI 10.1002/mrm.20493
Published online in Wiley InterScience (www.interscience.wiley.com).
© 2005 Wiley-Liss, Inc.
rately measure ADCs of water in anisotropic environments, a minimum of four individual DWMRI exams are
typically carried out: one without diffusion weighting and
three with diffusion weighting in orthogonal directions.
The diffusion-weighted images can then be geometrically
averaged to yield a “trace” image with average diffusion
weighting (2). Because the trace of the diffusion tensor is
invariant to rotation, the intensity in the resulting trace
image is invariant to tissue orientation (7). A number of
schemes for imparting isotropic diffusion weighting in
individual DWMRI scans have been proposed (8). In the
present paper we demonstrate a method that takes advantage of unique features of radial MRI data acquisition to
achieve this purpose.
In radial MRI, Fourier data are collected along radial
lines that all pass through the center of Fourier space. Each
line, therefore, contributes more or less equally to the
contrast in the final reconstructed image (9,10). If individual radial lines, or views, have different weightings, then
images reconstructed from such data will have contrast
determined by the average signal intensity from the different weightings. In radial fast spin-echo (radial-FSE) and
radial gradient and spin-echo, for example, individual radial lines are collected at different TE times within spin
echoes or gradient echoes and therefore have different T2
and/or T2* weighting (11). While this variation can introduce artifacts in reconstructed images, it can also be exploited to produce multiple images with variable contrast
from single radial data sets (11–13). In a diffusionweighted radial-FSE sequence, this feature can be exploited to achieve effectively isotropic diffusion weighting
within a single radial k-space data set, thereby increasing
the imaging speed of multishot diffusion-weighted radial
MRI.
In previous work, we have demonstrated a diffusionweighted radial-FSE sequence for obtaining diffusionweighted images with high spatial resolution and without
significant artifacts due to motion and/or susceptibility (14 –
16). In this method, diffusion weighting is achieved in a
preparation period where diffusion gradients are typically
turned on in a constant direction while acquiring a full radial
data set. This multishot method can yield higher spatial
resolution than conventional single-shot techniques; however, it requires more imaging time. In the present work, we
demonstrate a diffusion-weighted radial-FSE method for producing isotropic diffusion-weighted images from a single
radial k-space data set by altering the direction of diffusion
weighting gradients between TR periods. By selecting the
appropriate diffusion weighting directions and view ordering, images can be generated that have effectively isotropic
diffusion weighting without significant artifacts due to T2
decay, motion, and/or diffusion anisotropy.
1347
1348
Sarlls et al.
METHODS
An eddy current compensated diffusion-weighted radialFSE sequence is shown in Fig. 1. The preparation period
consists of CHESS presaturation of off-resonance fat, followed by a dual spin-echo (TE1) in which eddy-current
compensated diffusion gradients are played out (17). We
have found that this preparation works well for suppressing artifacts due to eddy currents without sacrificing significant signal from the small increase in time needed to
accommodate the second 180° RF pulse. Alternatively,
simple Stejskal–Tanner diffusion weighting can be incorporated within a single spin echo preparation. Following
this preparation, multiple views of Fourier data are acquired within a single TR period using a train of 180° RF
refocusing pulses with an echo spacing of TE2. Within
each data acquisition period, a full radial line is acquired
by moving out to the end of a half radial line, collecting
data while scanning back along a full radial line to the
opposite side of k-space, and returning to the origin. The
view ordering within the sequence is completely flexible
for this trajectory and is chosen to minimize the effects of
T2 decay and motion artifacts (16). In general, the full 2␲
radians of Fourier space should be coarsely sampled
within each echo train, i.e., each TR period, while differences in TE times for adjacent views are maximized. Artifacts due to motion are minimized by distributing motion
errors about the full 2␲ radians of Fourier space instead of
restricting them to sequential angular sections. This view
ordering also reduces artifacts due to T2 decay by spreading signal decay azimuthally with high angular frequency.
The general expression for determining view angles, ␪i,j,
within an optimal view ordering is
␪ i,j ⫽ [(i 䡠 ⌬␪echo ⫹ mi 䡠 ⌬␪) ⫹ j 䡠 ETL 䡠 ⌬␪] mod ␲
where i is the echo index, j is the echo train index, echo
train length (ETL) is the number of radial lines acquired
each excitation, ⌬␪echo ⫽ ␲/ETL, and the angular separation between consecutive radial lines, ⌬␪ ⫽ ␲/N, where N
is the total number of views collected. The i 䡠⌬␪echo term
forces the full 2␲ radians of Fourier space to be coarsely
sampled every TR period. The mi 䡠 ⌬␪ term is responsible
for mixing the TE times for adjacent views.
In isotropic diffusion-weighted radial-FSE, the diffusion
weighting direction is varied during the acquisition of a
FIG. 2. View orientation for the first two TR periods of isotropic
diffusion-weighted radial-FSE for a 256 view, ETL ⫽ 4 exam. The
view labels are (echo train number, echo number).
full radial k-space data set such that images reconstructed
from the data have effectively isotropic weighting. One
practical approach steps through the set of directions
[1,1,1], [⫺1,1,1], [1,⫺1,1], [⫺1,⫺1,1] (indicating full positive or negative gradient strength along the [X, Y, Z] axes)
during the scan. When arithmetically summed, this variation yields effectively isotropic weighting. By using fullstrength gradients on each of the three gradient coils simultaneously, a desired diffusion weighting (b-value) can
be obtained in the shortest amount of time (shortest TE).
To reduce artifacts that could arise from anisotropic diffusion, adjacent views must be collected with different diffusion directions. Therefore, to reduce artifacts due to
motion, T2 decay, and diffusion anisotropy, view ordering
must be carried out that (i) spreads radial lines acquired
within a given TR period about the full 2␲ radians of
Fourier space, (ii) maintains a high frequency variation of
TE with view angle, (iii) maintains a high frequency variation of diffusion weighting direction with view angle, and
(iv) does not correlate diffusion direction with TE. By
using the view ordering described previously (16) and
turning on diffusion weighting in each of the four directions for one-fourth the total TR periods consecutively, the
data acquisition satisfies conditions i, ii, and iii. To satisfy
condition iv, one more step had to be added to the view
order calculation. After calculating the array of view angles, ␪i,j, the views within a TR period undergo a circular
left shift with respect to the echo index i. The view angles
used in data acquisition ␪i⬘,j are then defined by
␪ i⬘, j ⫽ ␪ CSL(i, j mod ETL), j
FIG. 1. Diagram of the isotropic diffusion-weighted radial-FSE
pulse sequence with diffusion weighting in the XYZ direction and an
ETL ⫽ 4.
where CSL(i,j) defines a circular shift to the left of index i
by an amount j mod ETL. This view ordering results in the
relationships described graphically in Figs. 2 and 3. In Fig.
2, the views obtained in the first two TR periods of a 256
view, ETL ⫽ 4 acquisition are shown. Within each TR
Isotropic Diffusion-Weighted Radial Fast Spin-Echo MRI
FIG. 3. Echo number versus view number for each of the four
diffusion directions implemented in a 256 view, ETL ⫽ 4 isotropic
diffusion-weighted radial-FSE exam. The diffusion weighting direction is given in each panel. Note the high frequency variation of both
the echo position and the diffusion direction with view angle.
period, the entire range of view angles is coarsely sampled.
The relationship among TE, diffusion weighting direction,
and view angle is shown in Fig. 3. As can be seen, there is
a high frequency variation of both TE and diffusion
weighting direction with view angle and each diffusion
weighting direction is collected at a variety of uncorrelated
echo positions.
The isotropic diffusion-weighted radial-FSE sequence
was implemented on GE Signa 1.5 T LX echospeed and GE
Signa 3T MRI scanners (General Electric Medical Systems,
Milwaukee, WI, USA) with actively shielded gradients
capable of 33 and 40 mT/m, respectively. Image reconstruction of radial data sets was carried out offline using a
magnitude filtered backprojection reconstruction (FBPR)
implemented in IDL (Research Systems, Inc.) as described
previously (18,19).
RESULTS
Diffusion-weighted images from the brain of healthy volunteers were obtained using the conventional and isotro-
1349
pic diffusion-weighted radial-FSE methods. Representative images are shown in Fig. 4. Images collected with
constant diffusion weighting in four individual directions
within four individual exams are shown in Fig. 4a– d. The
variation in signal intensity of white matter structures in
the individual images indicates the anisotropic nature of
the tissue. A trace image obtained from the geometric
mean of these four images is shown in Fig. 4e and represents true isotropic diffusion weighting. An image of the
same slice obtained with isotropic diffusion weighting in a
single k-space data set is shown in Fig. 4f. Four identical
isotropic scans were averaged to obtain an equivalent signal-to-noise ratio (SNR) for comparison to the trace image.
The trace and isotropic diffusion-weighted radial-FSE images are qualitatively similar, indicating an effective averaging of anisotropic structures within the isotropic diffusion-weighted radial-FSE exam. The mean signal intensity
from a region of interest (ROI) in the left splenium of the
corpus callosum from each of the images in Fig. 4 is
plotted in Fig. 5. While the signal intensity in 4 a– d varies
significantly, the trace image and isotropic image have
comparable mean signal intensity: 245.1 (⫾ 23.9) and
250.8 (⫾ 18.8) in the trace and isotropic images, respectively. Other anisotropic structures in the brain exhibit
similar properties.
Conventional and isotropic diffusion-weighted radialFSE was also carried out in patients undergoing DWMRI
for acute stroke. Diffusion-weighted images of a stroke
patient obtained using conventional and isotropic diffusion-weighted radial-FSE methods are shown in Fig. 6.
Images obtained with constant diffusion weighting in four
individual directions in four individual exams are shown
in Fig. 6a– d. Figure 6e is the trace image produced from
the geometric mean of these four images. The corresponding isotropic diffusion-weighted radial-FSE image is
shown in Fig. 6f; this is from a single data set (no signal
averaging) and thus it has approximately half the SNR of
Fig. 6e, but required one-quarter of the time to acquire. The
region of stroke can be clearly seen in both images without
confounding hyperintensity from normal anisotropic
white matter. The tissue affected by the stroke appears
bright due to the reduced ADC associated with acute ischemia. The values of ADC in the region of the stroke obtained from ADC maps produced from the trace and isotropic images are 0.82 ⫻ 10⫺3 and 0.89 ⫻ 10⫺3 mm2/s,
respectively. This can be compared to the ADC in normal
white mater (1.28 ⫻ 10⫺3 and 1.31 ⫻ 10⫺3 mm2/s for the
trace and isotropic exams, respectively), indicating a drop
in ADC of approximately 34%, consistent with acute
stroke. The isotropic radial-FSE method not only yields
qualitatively equivalent images as the conventional
method, but also quantitatively maintains all aspects of the
areas of ischemia, regardless of the lower SNR.
DISCUSSION
A unique feature of all radial MRI methodologies is that
the center of Fourier space is sampled by each radial line
acquired. In the present paper, we have demonstrated that
this feature enables high-resolution isotropic diffusionweighted images to be obtained from a single radial kspace data set. While isotropic diffusion-weighted radial-
1350
Sarlls et al.
FIG. 4. Diffusion-weighted radial-FSE images of a normal volunteer at 1.5 T. Parameters for all
images: b ⫽ 1000 s/mm2, TE1 ⫽
70 ms, TE2 ⫽ 13 ms, TR ⫽
1500 ms, ETL ⫽ 4, FOV ⫽ 26 ⫻
26 cm2, with slice thickness ⫽ 5
mm. Images obtained with diffusion weighting in the XYZ, ⫺XYZ,
X⫺YZ, and ⫺X⫺YZ directions are
shown in a, b, c, and d, respectively. A trace image produced
from the geometric mean of these
images is shown in e. An isotropic
diffusion-weighted radial-FSE image is shown in f. This image was
produced from the average of
four identical exams so that signal
to noise is comparable to that in
e.
FSE produces images that have effectively isotropic diffusion weighting, they are not identical to true trace images
produced from the average of individual data sets. When a
trace image is produced from the geometric mean of n
individual scans obtained with individual diffusion
weighting, the resulting signal Strace has the average diffusion weighting of the individual scans (S1,2. . . n) and can be
expressed as
S trace ⫽
冑nS1 ⫻ S2. . . ⫻ Sn
n
⫽冑
S0e⫺bD1 ⫻ S0e⫺bD2. . . ⫻ S0e⫺bDn
n ⫺b共D 1 ⫹D 2 . . .⫹D n 兲
n ⫺nbDave
⫽ S0 冑
e
⫽ S0 冑
e
⫽ S0e⫺bDave,
where it is assumed that k appropriate diffusion directions
were selected such that
D ave ⫽
1
n
冘
n
Dk.
k⫽1
The FBPR process used in reconstruction of radial data
sets, however, is a linear operation and results in an arithmetic averaging of the signal obtained from the set of
diffusion directions chosen. Additionally, there is an individual point-spread function (PSF) associated with the
angular undersampling of each diffusion direction. The
signal in an isotropic diffusion-weighted radial-FSE image, Siso, can be written as
S iso ⫽
S1 ⫹ S2 . . . ⫹ Sn
n
⫽
S0e⫺bD 1 * psf1 ⫹ S0e⫺bD 2 * psf2 . . . ⫹ S0e⫺bD n * psfn
n
⫽
S0 ⫺bD 1
* psf1 ⫹ e⫺bD 2 * psf2 . . . ⫹ e⫺bD n * psfn).
(e
n
If diffusion in the tissue is isotropic, then D1 ⫽ D2 ⫽ Dn ⫽
Dave and psf1 ⫹ psf2 . . . ⫹ psfn ⫽ ␦ (delta function), leaving
FIG. 5. The mean signal intensity for identical ROIs in the left
splenium of the corpus callosum from images in Fig. 4; the black bar
indicates the noise level in each image.
S iso ⫽
S0 ⫺bD ave
) ⫽ S0e⫺bD ave.
(ne
n
Isotropic Diffusion-Weighted Radial Fast Spin-Echo MRI
1351
FIG. 6. Diffusion-weighted radial-FSE images of a stroke patient acquired at 1.5 T. Parameters for all images: b ⫽ 1000
s/mm2, TE1 ⫽ 70 ms, TE2 ⫽
13 ms, TR ⫽ 1000 ms, ETL ⫽ 4,
FOV ⫽ 26 ⫻ 26 cm2, with slice
thickness ⫽ 5 mm. Images obtained with diffusion weighting in
the XYZ, ⫺XYZ, X⫺YZ, and
⫺X⫺YZ directions are shown in
a, b, c, and d, respectively. A
trace image produced from the
geometric mean of these images
is shown in e. The corresponding
isotropic diffusion-weighted radial-FSE image is shown in f.
If the tissue is anisotropic, however, the true average diffusion-weighted signal will be multiplied by a small factor, which is always positive, making the signal in the
isotropic diffusion-weighted radial-FSE image slightly
higher than that of the trace image. This relationship is
shown graphically in Fig. 7. Calculated signal intensity
from three diffusion-weighted data sets, with a b-value ⫽
1000 s/mm2, using arithmetic (dashed line) and geometric
(solid line) averaging, are plotted versus fractional anisotropy (FA). The mean diffusivity was chosen to be 1.0 ⫻
10⫺3 mm2/s to represent white matter. The relationship
between the diffusion coefficient for the three orthogonal
directions range from equal values (FA ⫽ 0) to diffusion in
only one direction (FA ⫽ 1.0). The error produced from the
FIG. 7. Signal intensity versus fractional anisotropy expected from
three orthogonal diffusion-weighted data sets using arithmetic
(black dashed line) or geometric (black solid line) averaging. The
data were produced assuming b ⫽ 1000 s/mm2 and mean diffusivity
of 1.0 ⫻ 10⫺3 mm2/s. The relationship between the diffusion coefficients for the three orthogonal directions ranges from equal values
(FA ⫽ 0) to diffusion in only one direction (FA ⫽ 1.0). The error
produced from the arithmetic averaging is shown as a percentage of
the geometric mean (gray solid line).
arithmetic averaging is shown in Fig. 7 as a percentage of
the geometric mean. For isotropic structures, the two calculations are identical. At FA ⫽ 0.8, an appropriate value
for highly anisotropic white matter, the error is approximately 6.6%. Such small error is consistent with the experimental results shown in Figs. 4 and 5.
Due to the motion insensitivity of radial MRI, reconstructed radial-FSE images are free from significant motion
artifacts and due to the fast spin-echo refocusing utilized
in the sequence, images have no susceptibility artifacts as
in conventional DWMRI methodology. Figure 8a is a diffusion-weighted trace image obtained with single-shot
echo planar imaging (SSEPI). Three images obtained with
diffusion weighting in three orthogonal directions were
averaged to obtain the trace image. An isotropic diffusionweighted radial-FSE image of the same patient is shown in
Fig. 8b. In both images, areas of ischemia are clearly visible
as regions of hyperintensity, there is no obscuring bright
signal due to diffusion anisotropy in white matter, and
there are no artifacts due to motion. However, the lower
resolution SSEPI method is unable to resolve the two
smaller anterior areas of ischemia, seen in the higher resolution radial-FSE image. The lack of image distortion in
the radial-FSE image also allows for a more direct comparison of diffusion-weighted images to conventional T2weighted and T1-weighted images. Similar results have
been obtained consistently in our clinical evaluations.
Because radial-FSE is insensitive to magnetic field inhomogeneity, it works well at higher field strengths. A comparison of isotropic diffusion-weighted radial-FSE with
SSEPI at 3 T is shown in Fig. 9. These images are from a
patient with a renal cell carcinoma metastasis within the
right hemisphere and were obtained after the injection of a
contrast agent, Omniscan, which has been shown to have
insignificant effects on DWMRI (20). DWMRI is being used
to evaluate the cellularity of cancerous tissues (21) and to
predict tumor response to chemotherapy (3,22–24). The
SSEPI images in Fig. 9 have 3.75 ⫻ 3.75 mm2 in plane
resolution and show typical susceptibility artifacts in the
frontal and temporal lobe regions due to air-tissue inter-
1352
Sarlls et al.
FIG. 8. Diffusion-weighted images acquired on
the 1.5 T scanner of a stroke patient (b ⫽
1000 s/mm2) using SSEPI (a) and isotropic diffusion-weighted radial-FSE (b). For the SSEPI exam
TE ⫽ 81 ms, TR ⫽ 10 s, FOV ⫽ 30 ⫻ 20 cm2, with
slice thickness ⫽ 5 mm. For the radial-FSE exam,
TE1 ⫽ 70 ms, TE2 ⫽ 13 ms, TR ⫽ 1500 ms, ETL ⫽
4, FOV ⫽ 26 ⫻ 26 cm2, with slice thickness ⫽
5 mm.
faces. No detail is visible within or adjacent to the tumor.
The isotropic diffusion-weighted radial-FSE images
shown in Fig. 9c and d have 0.94 ⫻ 0.94 mm2 in-plane
resolution and are free of motion and susceptibility artifacts. Due to the high resolution of the radial-FSE images
and insensitivity to susceptibility changes, heterogeneity
within and around the tumor can be clearly visualized
(Fig. 9c). Because this metastasis was hemorrhagic, it contains blood products that yield regions of local susceptibility gradients that cause problems for SSEPI methods.
A unique feature of isotropic diffusion-weighted radialFSE is that both ADC maps and T2 maps can be generated
from only two data sets. A high-resolution ADC map,
obtained from two individual exams (b ⫽ 5 s/mm2 and b ⫽
1000 s/mm2 in Fig. 10a and b, respectively) is shown in
Fig. 10c. ADC values in the brain unaffected by the stroke
are similar to those given in literature (1.17 ⫻ 10⫺3 and
1.32 ⫻ 10⫺3 mm2/s for white matter and gray matter,
respectively) and there is a decrease in the ADC in the
region of ischemia (0.72 ⫻ 10⫺3 mm2/s versus 0.96 ⫻ 10⫺3
FIG. 9. Diffusion-weighted images (b ⫽ 1000 s/mm2)
of a cancer patient acquired at 3 T using SSEPI
(a,b) and isotropic diffusion-weighted radial-FSE
(c,d). Imaging parameters for the SSEPI exam were
TE ⫽ 72 ms, TR ⫽ 10 s, FOV ⫽ 30 ⫻ 20 cm2, and
slice thickness ⫽ 5 mm. Imaging parameters for
the radial-FSE exam were TE1 ⫽ 60 ms, TE2 ⫽
15 ms, TR ⫽ 3000 ms, ETL ⫽ 4, FOV ⫽ 26 ⫻
26 cm2, with slice thickness ⫽ 5 mm.
mm2/s in homologous tissue on the contralateral side)
(25,26). It is important to note that measured ADC values
are dependent on the amount of diffusion weighting applied, the diffusion time, and the TE of the sequence (26).
ADC values in regions of ischemia are also dependent on
the time elapsed between stroke onset and the DWMRI
exam. Maps of T2 can be calculated from any individual
isotropic radial-FSE data set. An example of this is shown
in Fig. 10d where a T2 map was generated from the nondiffusion-weighted (b ⫽ 5 s/mm2) isotropic radial-FSE
data set (Fig. 10a), with a postprocessing technique described previously (12,13). Briefly, multiple images at different effective TE values are produced from one radialFSE data set. For reconstruction of an image at a particular
effective TE, only views acquired at that TE provide data
near the center of k-space (within the Nyquist radius). At
higher spatial frequencies, data from views with other TEs
are included (13). A T2 map is then calculated from the TE
images using a single exponential decay fit. There is an
increase of T2 in the region of ischemia (139.8 ms versus
Isotropic Diffusion-Weighted Radial Fast Spin-Echo MRI
1353
FIG. 10. Images acquired on the 1.5 T scanner of
a stroke patient using isotropic diffusion-weighted
radial-FSE. For all images TE1 ⫽ 70 ms, TE2 ⫽
13 ms, TR ⫽ 1500 ms, ETL ⫽ 4, FOV ⫽ 26 ⫻
26 cm2, slice thickness ⫽ 5 mm. (a) is the b ⫽
5 s/mm2 image, and (b) is the b ⫽ 1000 s/mm2
image. An ADC map calculated from the radialFSE data set is shown in (c). A T2 map calculated
from the radial-FSE b ⫽ 5 s/mm2 image is shown
in (d).
110.2 ms in the contralateral side). This region of ischemia
shows a decreased ADC with an elevated T2, consistent
with acute stroke imaged between 6 and 48 hr (1,27). The
ability to quantitatively determine both ADC and T2 in
ischemic tissue may help determine the age of the stroke,
which plays an important role in therapeutic decisions
(28).
Other methods have been developed to obtain effectively isotropic diffusion weighting within the preparation
period of a sequence (8). A common feature of these methods is that they require an increased amount of time to
achieve a given diffusion weighting (compared to Stejskal–
Tanner type weighting). Because the diffusion-preparation
periods are then required to be longer, the minimum TE
time increases, resulting in a loss of SNR. Also, it is not
always possible to make such sequences eddy current
compensated, which we have found to be very helpful in
both radial and Cartesian DWMRI. Because subsets of the
radial data are acquired with diffusion weighting in different directions, it may be possible to obtain information on
diffusion anisotropy within a single radial data set (29).
This is similar to producing T2 maps from individual
radial-FSE data sets as described previously (12,13) and is
currently being investigated.
CONCLUSIONS
The present paper demonstrates a simple method for obtaining isotropic diffusion weighting in a single radial MRI
exam. Using the appropriate view ordering and diffusion
weighting, radial-FSE images can be acquired with high
spatial resolution that are insensitive to bulk motion, susceptibility changes, T2 decay, and diffusion anisotropy.
Radial-FSE also allows for T2 maps to be generated from
individual exams such that both ADC and T2 maps can be
obtained from only two radial-FSE exams.
REFERENCES
1. Warach S, Gaa J, Siewert B, Wielopolski P, Edelman RR. Acute human
stroke studied by whole brain echo planar diffusion-weighted magnetic
resonance imaging. Ann Neurol 1995;37:231–241.
2. van Gelderen P, de Vleeschouwer M, DesPres D, Pekar J, van Zijl P,
Moonen C. Water diffusion and acute stroke. Magn Reson Med 1994;
31:154 –163.
3. Chenevert TL, McKeever PE, Ross BD. Monitoring early response of
experimental brain tumors to therapy using diffusion magnetic resonsance imaging. Clin Caner Res 1997;3:1457–1466.
4. Nakahara M, Ericson K, Bellander BM. Diffusion-weighted MR and
apparent diffusion coefficient in the evaluation of severe brain injury.
Acta Radiol 2001;42:365–369.
5. Bammer R, Fazekas F. Diffusion imaging in multiple sclerosis. Neuroimaging Clin North Am 2002;12:71–106.
6. Pierpaoli C, Basser PJ. Toward a quantitative assessment of diffusion
anisotropy. Magn Reson Med 1996;36:893–906.
7. Basser PJ, Mattiello J, Le Bihan D. MR diffusion tensor spectroscopy
and imaging. Biophys J 1994;66:259 –267.
8. Wong E, Cox R, Song A. Optimized isotropic diffusion weighting. Magn
Reson Med 1995;34:139 –143.
9. Hall LD, Sukumar S. Rapid data-acquisition for NMR imaging by projection-reconstruction method. J Magn Reson 1984;56:179 –182.
1354
10. Rasche V, Holz D, Schepper W. Radial turbo spin echo imaging. Magn
Reson Med 1994;2:629 – 638.
11. Gmitro AF, Kono M, Theilmann RJ, Altbach MI, Li Z, Trouard TP.
Radial GRASE: Implementation and Applications. Magn Reson Med
2005;53:1363–1371.
12. Song HK, Dougherty L. k-Space weighted image contrast (KWIC) for
contrast manipulation in projection reconstruction MRI. Magn Reson
Med 2000;44:825– 832.
13. Altbach MI, Outwater EK, Trouard TP, Krupinski EA, Theilmann RJ,
Stopeck AT, Kono M, Gmitro AF. Radial fast spin-echo method for
T2-weighted imaging and T2 mapping of the liver. J Magn Reson
Imaging 2002;16:179 –189.
14. Trouard TP, Theilmann RJ, Altbach MI, Gmitro AF. High-resolution
diffusion imaging with DIFRAD-FSE (diffusion-weighted radial acquisition of data with fast spin-echo) MRI. Magn Reson Med 1999;42:11–
18.
15. Trouard TP, Saul J, Seeger J, Theilmann RJ, Altbach MI, Gabaeff D,
Gmitro AF. Comparison of DIFRAD-FSE with single-shot EPI for diffusion-weighted MRI of acute stroke. In: Proceedings of the 10th Annual
Meeting of ISMRM, Honolulu, 2002. p 1114.
16. Theilmann RJ, Gmitro AF, Altbach MI, Trouard TP. View-ordering in
radial fast spin-echo imaging. Magn Reson Med 2004;51:768 –774.
17. Reese TG, Heid O, Weisskoff RM, Wedeen. Reduction of eddy-currentinduced distortion in diffusion MRI using a twice-refocused spin echo.
Magn Reson Med 2003;49:177–182.
18. Gmitro AF, Alexander AL. Use of projection reconstruction imaging
method to decrease motion sensitivity in diffusion-weighted MRI.
Magn Reson Med 1993;29:835– 838.
19. Jung K, Cho ZH. Reduction of flow artifacts in NMR diffusion imaging
using view-angle tilted line-integral projection reconstruction. Magn
Reson Med 1991;19:249 –293.
20. Yamada K, Kubota H, Kizu O, Nakamura H, Ito H, Yuen S, Tanaka O,
Kubota T, Makino M, Van Cauteren M, Nishimura T. Effect of intravenous Gadolinium-DTPA on diffusion-weighted images, evaluation of
normal brain and infarcts. Stroke 2002;33:1799 –1802.
Sarlls et al.
21. Sugahara T, Korogi Y, Kochi M, Ikushima I, Shigematu Y, Hirai T,
Okuda T, Liang L, Ge Y, Komohara Y, Ushio Y, Takahashi M. Usefulness of diffusion-weighted MRI with echo-planar technique in the
evaluation of cellularity in gliomas. J Magn Reson 1999;9:53– 60.
22. Galons JP, Altbach MI, Paine-Murrieta GD, Taylor CW, Gillies RJ. Early
increases in breast tumor xenograft water mobility in response to Paclitaxel therapy detected by non-invasive diffusion magnetic resonance
imaging. Neoplasia 1999;1:113–117.
23. Zhao M, Pipe JG, Bonnett J, Evelhoch JL. Early detection of treatment
response by diffusion-weighted H-NMR spectroscopy in a murine tumour in vivo. Br J Cancer 1996;73:61– 64.
24. Jennings D, Hatton BN, Guo J, Galons JP, Trouard TP, Raghunand N,
Marshall J, Gillies RJ. Early response of prostate carcinoma xenografts
to docetaxel chemotherapy monitored with diffuson MRI. Neoplasia
2002;4:255–262.
25. Sorensen AG, Buonano FS, Gonzalez RG, Schwamm LH, Lev MH,
Huang-Hellinger FRH, Reese TG, Weisskoff RM, Davis TL, Suwanwela
N, Can U, Moreira JA, Copen WA, Look RB, Finklestein SP, Rosen BR,
Koroshetz WJ. Hyperacute stroke: evaluation with combined multislice
diffusion-weighted and hemodynamically weighted echo-planar MR
imaging. Radiology 1996;199:391– 401.
26. Yoshiura T, Wu O, Zaheer A, Reese TG, Sorensen AG. Highly diffusionsensitized MRI of brain: Dissociation of gray and white matter. Magn
Reson Med 2001;45:734 –740.
27. Welch KMA, Windham J, Knight RA, Nagesh V, Hugg JW, Jacobs M,
Peck D, Booker P, Dereski MO, Levine SR. A model to predict the
histopathology of human stroke using diffusion and T2-weighted magnetic resonance imaging. Stroke 1995;26:1983–1989.
28. Eastwood JD, Engelter ST, MacFall JF, Delong DM, Provenzale JM.
Quantitative assessment of the time course of infarct signal intensity on
diffusion-weighted images. Am J Neuroradiol 2003;24:680 – 687.
29. Newbould RD, Sarlls JE, Trouard TP. Data collection and postprocessing strategies in radial-FSE for diffusion tensor analysis. In:
Proceedings of the 12th Annual Meeting of ISMRM, Kyoto, Japan,
2004. p 446.